Semantic Network Interpretation
This work addresses the problem of interpreting neural networks for researchers and practitioners by providing semantic insights as an alternative to visualization, though it is incremental in building on existing interpretation methods.
The paper tackles semantic network interpretation by representing filter concepts with visual attribute distributions and generating explanatory sentences for decisions, using a Bayesian inference algorithm to link filters and decisions to attributes, with human studies confirming its benefits and demonstrating its role in understanding failure patterns and model performance correlations.
Network interpretation as an effort to reveal the features learned by a network remains largely visualization-based. In this paper, our goal is to tackle semantic network interpretation at both filter and decision level. For filter-level interpretation, we represent the concepts a filter encodes with a probability distribution of visual attributes. The decision-level interpretation is achieved by textual summarization that generates an explanatory sentence containing clues behind a network's decision. A Bayesian inference algorithm is proposed to automatically associate filters and network decisions with visual attributes. Human study confirms that the semantic interpretation is a beneficial alternative or complement to visualization methods. We demonstrate the crucial role that semantic network interpretation can play in understanding a network's failure patterns. More importantly, semantic network interpretation enables a better understanding of the correlation between a model's performance and its distribution metrics like filter selectivity and concept sparseness.